Quantitative characterization and micro-CT mineral mapping of natural fissural enamel lesions

Abstract

Objectives

The aim of this study was to characterize the mineral distribution pattern of natural fissural enamel lesions and to quantify structural parameters and mineral density of these lesions in comparison to proximal white spot enamel lesions.

Methods

Imaging was undertaken using a high-resolution desktop micro-computed tomography system. A calibration equation was used to transform the grey level values of images into true mineral density values. The value of lesion parameters including the mineral density and the thickness of the surface layer of the enamel lesion were extracted from mineral density profiles.

Results

The thickness of the surface layer showed variation among different lesions and it ranged from 0–90 μm in proximal lesions and 0–137 μm in fissural lesions. The average thickness of surface layer in fissural lesions was significantly higher than smooth surface proximal lesions. Sound fissural enamel showed lower mineral density compared to proximal enamel.

Conclusion

Micro-CT and the suggested de-noising and visualization method provide an efficient high-resolution approach for non-destructive evaluation of fissural lesions. Using these methods, the current study revealed the exclusive pattern and structure of fissural enamel lesions which may provide a basis for future studies on prevention and remineralization of these lesions.

Clinical significance

The common demineralization pattern of fissural lesions, which indicates the extension of the lesion in two directions towards the pulp horns, may explain the early inflammation and symptoms of the pulp in fissural lesions even when the lesion base appears far from the pulp roof in normal radiographs. In addition, the presence of the surface layer, indicates that vigorous probing of the occlusal fissures may lead to breakage and cavitation of the enamel lesions.

Introduction

During evolution, the dentition of omnivores evolved into a heterodont system comprising of specialized teeth with various morphologies and complex occlusal surfaces. This transition improved the efficacy of their masticatory function and enhanced the utilization of food caloric energy for more intricate physical and intellectual activities. Accordingly, human teeth attained multi-cuspid crown morphology with pointed occlusal cusps and ridges which are separated by grooves and fissures. The multi-cuspid morphology not only improved the masticatory and nutritional functions of the human teeth, but also created a protected and stable occlusion through a balanced cusp fossa system.

From an embryologic perspective, the genesis of developmental grooves and the determination of the morphology of the dental crown occur during the bell stage of the highly coordinated odontogenesis process. The grooves are formed following the folding invagination of the inner enamel epithelium and the coalescence between the adjacent developmental lobes . While the normal coalescence of developing lobes forms shallow grooves and fossae, their abnormal and incomplete fusion creates deep and narrow pits and fissures on the tooth occlusal surface.

In spite of the remarkable functional and structural benefits of the multi-cuspid dentition, the complex morphology of the dental occlusal surface and the presence of hygienically inaccessible pits and fissures have made the occlusal fissure, a potential site for the retention and maturation of microbial plaque. Accordingly, the occlusal surfaces of teeth, particularly of the first and second molars, are the most susceptible areas to caries attack and have the shortest survival times among all teeth surfaces . Presently, fissure caries are the most common type of carious lesions in both primary and permanent dentition.

In addition, the obscured form of the occlusal fissures creates challenges for clinical detection and investigational study of fissural lesions. Therefore, despite the existing extensive knowledge about smooth surface enamel caries , little is known about the internal structure and mineral distribution pattern of carious fissural lesions. Many of the common study methods such as transverse microradiography , scanning electron microscopy and polarized light microscopy require cutting and sectioning of the enamel lesion specimens. Since the exact position and form of the fissural lesions are not directly visible, it is not easy to obtain the appropriate study sections. Moreover, the fine features of the fissural lesions such as enamel surface layer can undergo damage during the sectioning process. Therefore non-invasive three dimensional methods such as micro-CT are required for the study of natural fissural enamel lesions.

Micro-CT is an X-ray attenuation method which provides high resolution three dimensional information of the internal structure of hard tissues and materials. Using this method, volumetric X-ray attenuation measurements reveal the spatial distribution of attenuation coefficients ( μ ) within the object and without the superimposition of other structures. Virtual study sections in any desired direction can be generated using the reconstructed 3-dimensonal volume of the scanned object. The high level of sensitivity and specificity of micro-CT in relation to the gold standard methods of histology and transverse micro radiography have been confirmed by correlative studies .

While micro-CT has been used for the study of different types of dental lesions including artificial caries , natural proximal enamel lesions and hypomineralized enamel , this approach has not been utilized for the systematic study of fissural enamel lesions in human dentition. Considering the high rate of fissural caries and the current need for developing caries diagnostic and curative methods, further investigation is required to elucidate the structure, mineral patterns and formation mechanisms of fissural lesions. Therefore the aim of this study was to characterize the mineral distribution pattern of natural fissural enamel lesions and to quantify structural parameters and mineral density of these lesions in comparison to proximal white spot enamel lesions.

Materials and methods

Study specimens

Human molar and premolar teeth were collected from oral surgery and orthodontics departments at Sydney Dental Hospital, The University of Sydney, with the approval from Sydney local health district ethics review committee, protocol No X12–0065 & HREC/12/RPAH/106.

Extracted teeth were partly sterilized in Milton anti-bacterial solution (0.95% sodium hypochlorite, Milton Australia PTY LTD.) for 30 min and then brushed clean under running water to remove any plaque. Following this decontamination process, the samples were stored in Hanks Balanced Salt Solution (HBSS) at 4 °C prior to use. Thymol granules (Sigma–Aldrich, Australia) were added to the solution for disinfection and prevention of fungal growth. Teeth were visually assessed by two experienced clinicians and non-cavitated teeth were selected for micro-CT scanning. Following the establishment of the diagnosis, twenty fissural lesions (ICDAS 1 or 2) and twenty proximal white spot lesions (ICDAS 1 or 2) were selected as study subjects. The control groups included twenty sound fissural (ICDAS 0) and twenty sound proximal enamel areas (ICDAS 0). In sound specimens, there was no evidence of carious activity in enamel including carious discoloration or carious opacity after air drying. Teeth with extrinsic or intrinsic stains were considered as sound.

For the X-ray micro-CT scanning, each specimen was rigidly fixed inside a plastic tube which was itself glued to a metal mount for attachment to the XRMT rotation stage. The samples were wrapped in Parafilm ‘M’ (American National Can, Chicago, IL, USA) to minimize drying during the scanning process.

X-ray micro tomography scanning parameters and tomographic reconstruction

Imaging was undertaken using a high resolution desktop micro-computed tomography system (Skyscan 1172, Skyscan, N.V, Aartsellar, Belgium). With the SkyScan 1172, a fourth generation compact desktop machine, X-rays are generated by a tungsten anode in a sealed micro-focus X-ray tube, and the incident X-ray photons are converted to light by a scintillator. The light is detected by a two-dimensional (2000 pixels × 1048 pixels) CCD-camera which is connected to a frame-grabber.

Imaging of the teeth and phantoms was undertaken at an accelerating source voltage of 100 keV, a source current of 100 μA and an exposure time of 885 ms. An inbuilt filter composed of 1.0 mm thickness of aluminum and 0.05 mm of copper was applied to remove low energy X-rays and to restrict X-rays spectral bandwidth of the polychromatic radiation. The equivalent monochromatic energy spectrum of filtered X-ray had an effective mean energy of 60 keV. The long axes of the teeth were parallel with the center of rotation of the mounting device. The samples were rotated over 360° at angular increments of 0.14° producing 2570 projections with an image matrix of 2000 pixels × 1048 pixels.

Grey level calibration was achieved using three hydroxyapatite discs with known low, medium and high mineral densities in order to obtain the mineral density values of different parts of the tooth . These three hydroxyapatite discs were positioned and fixed on top of each tooth before starting the imaging session. Details for the fabrication process and the materials used for constructing the phantoms are described elsewhere .

At the end of the scanning process, the resultant 2-D shadow projection images ( Fig. 1 A) were saved as 16 bit Tagged Image File Format (TIFF) and were exported to a 3-D cone beam reconstruction program (NRecon software, version 1.4.4; SkyScan) which uses a modified Feldkamp cone-beam reconstruction algorithm to reconstruct the 3-D object. A value of 60 was chosen for correction of beam hardening artefact in the NRecon software.

Fig. 1
(A) 2-D shadow projection image of the tooth (B) a reconstructed slice view of the same tooth, at the position of yellow line, after tomographic reconstruction. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The tomographic reconstruction software produced a dataset of slice views in 16 bit tiff format, which were perpendicular to the specimen rotation axis and had a voxel size resolution of 8.82 μm ( Fig. 1 B). Vertical slice views were produced by re-slicing the reconstructed volume of the whole image stack in FIJI (W.S. Rasband, U.S. National Institutes of Health, Bethesda, Md, USA, , 1997–2011).

De-noising, visualization and quantification

Following the tomographic reconstruction, the stacks of images were consequently imported into image processing programs comprising FIJI (W.S. Rasband, U.S. National Institutes of Health, Bethesda, Md, USA, , 1997–2011) and MATLAB (MatLab R2012b 8.0.0.783, Mathworks, Natick, MA, USA).

Prior to visualization and analysis, the images needed to undergo de-noising to increase the signal to noise ratio and to remove the noise appearing as bright and dark points in all areas of the images. De-noising was performed using a developed algorithm based on total variation regularization , with the following parameters:

μ (regularization parameter) = 0/04

ρr (initial penalty parameter) = 3 and ρo = 40

In total variation de-noising, the goal is to reduce the high total variation of the noisy image and therefore to eliminate the undesired details while preserving the original information. The de-noising algorithm was compiled using MATLAB Version 7.11.0.584(R2010b) on a computer system with Intel ® Core™ i7CPU (Q740 @ 1.73 GHz), and 4.00 GB RAM memory.

For mineral density calibration, the grey level values of 10 points on selected images of each HA phantom, were measured, averaged and plotted against mineral density value of the phantoms in order to calculate the calibration equation which was as follows:

Mineral density = (Grey level value × 0.0159) + 0.2967

The calibration equation was used to transform the grey level values of the images into true mineral density values. Each lesion was color coded based on the grey level values and corresponding mineral densities to provide a mineral map of the lesion and render an enhanced visualization. The visualization and color coding were performed using colormapeditor command by choosing Jet color map in MATLAB. Fixed RGB (Red, Green, Blue) index values were set for all the colorized images.

The values of lesion parameters including the mineral density and the thickness of the surface layer were extracted from mineral density profiles plotted in FIJI and the visualized mineral maps in MATLAB.

For calculating the mineral density (MD) of each region, the MD values of all of the pixels in the region were measured and averaged. In the carious lesions, the regions of interest (ROIs) were selected at the surface layer as well as the center of the lesion body. In the sound specimens, the ROIs were selected at the outer layer as well as the inner layer of enamel.

The thickness of the surface layer of the carious lesions was measured using line scans of the mineral content profile across the sidewalls and the floor of the fissures and at the central traverse of the proximal lesions. The distance from the enamel surface up to the first reduction step in the mineral density profile indicating the beginning of the lesion body was measured on the density profiles. In cases where there was not an obvious reduction step in the mineral profile, the value was obtained by measuring the thickness of the color zone corresponding to the surface layer in the colorized images. The measurements were checked for normal distribution. The statistical analysis was performed using statistical software GraphPad Prism (Graphpad Software, San Diego, CA) to test for differences between the means. The results were analyzed by One-way analysis of variance (ANOVA). Multiple comparisons between groups were performed by posthoc Tukey test. P -values less than 0.05 were considered to be statistically significant.

Materials and methods

Study specimens

Human molar and premolar teeth were collected from oral surgery and orthodontics departments at Sydney Dental Hospital, The University of Sydney, with the approval from Sydney local health district ethics review committee, protocol No X12–0065 & HREC/12/RPAH/106.

Extracted teeth were partly sterilized in Milton anti-bacterial solution (0.95% sodium hypochlorite, Milton Australia PTY LTD.) for 30 min and then brushed clean under running water to remove any plaque. Following this decontamination process, the samples were stored in Hanks Balanced Salt Solution (HBSS) at 4 °C prior to use. Thymol granules (Sigma–Aldrich, Australia) were added to the solution for disinfection and prevention of fungal growth. Teeth were visually assessed by two experienced clinicians and non-cavitated teeth were selected for micro-CT scanning. Following the establishment of the diagnosis, twenty fissural lesions (ICDAS 1 or 2) and twenty proximal white spot lesions (ICDAS 1 or 2) were selected as study subjects. The control groups included twenty sound fissural (ICDAS 0) and twenty sound proximal enamel areas (ICDAS 0). In sound specimens, there was no evidence of carious activity in enamel including carious discoloration or carious opacity after air drying. Teeth with extrinsic or intrinsic stains were considered as sound.

For the X-ray micro-CT scanning, each specimen was rigidly fixed inside a plastic tube which was itself glued to a metal mount for attachment to the XRMT rotation stage. The samples were wrapped in Parafilm ‘M’ (American National Can, Chicago, IL, USA) to minimize drying during the scanning process.

X-ray micro tomography scanning parameters and tomographic reconstruction

Imaging was undertaken using a high resolution desktop micro-computed tomography system (Skyscan 1172, Skyscan, N.V, Aartsellar, Belgium). With the SkyScan 1172, a fourth generation compact desktop machine, X-rays are generated by a tungsten anode in a sealed micro-focus X-ray tube, and the incident X-ray photons are converted to light by a scintillator. The light is detected by a two-dimensional (2000 pixels × 1048 pixels) CCD-camera which is connected to a frame-grabber.

Imaging of the teeth and phantoms was undertaken at an accelerating source voltage of 100 keV, a source current of 100 μA and an exposure time of 885 ms. An inbuilt filter composed of 1.0 mm thickness of aluminum and 0.05 mm of copper was applied to remove low energy X-rays and to restrict X-rays spectral bandwidth of the polychromatic radiation. The equivalent monochromatic energy spectrum of filtered X-ray had an effective mean energy of 60 keV. The long axes of the teeth were parallel with the center of rotation of the mounting device. The samples were rotated over 360° at angular increments of 0.14° producing 2570 projections with an image matrix of 2000 pixels × 1048 pixels.

Grey level calibration was achieved using three hydroxyapatite discs with known low, medium and high mineral densities in order to obtain the mineral density values of different parts of the tooth . These three hydroxyapatite discs were positioned and fixed on top of each tooth before starting the imaging session. Details for the fabrication process and the materials used for constructing the phantoms are described elsewhere .

At the end of the scanning process, the resultant 2-D shadow projection images ( Fig. 1 A) were saved as 16 bit Tagged Image File Format (TIFF) and were exported to a 3-D cone beam reconstruction program (NRecon software, version 1.4.4; SkyScan) which uses a modified Feldkamp cone-beam reconstruction algorithm to reconstruct the 3-D object. A value of 60 was chosen for correction of beam hardening artefact in the NRecon software.

Fig. 1
(A) 2-D shadow projection image of the tooth (B) a reconstructed slice view of the same tooth, at the position of yellow line, after tomographic reconstruction. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The tomographic reconstruction software produced a dataset of slice views in 16 bit tiff format, which were perpendicular to the specimen rotation axis and had a voxel size resolution of 8.82 μm ( Fig. 1 B). Vertical slice views were produced by re-slicing the reconstructed volume of the whole image stack in FIJI (W.S. Rasband, U.S. National Institutes of Health, Bethesda, Md, USA, , 1997–2011).

De-noising, visualization and quantification

Following the tomographic reconstruction, the stacks of images were consequently imported into image processing programs comprising FIJI (W.S. Rasband, U.S. National Institutes of Health, Bethesda, Md, USA, , 1997–2011) and MATLAB (MatLab R2012b 8.0.0.783, Mathworks, Natick, MA, USA).

Prior to visualization and analysis, the images needed to undergo de-noising to increase the signal to noise ratio and to remove the noise appearing as bright and dark points in all areas of the images. De-noising was performed using a developed algorithm based on total variation regularization , with the following parameters:

μ (regularization parameter) = 0/04

ρr (initial penalty parameter) = 3 and ρo = 40

In total variation de-noising, the goal is to reduce the high total variation of the noisy image and therefore to eliminate the undesired details while preserving the original information. The de-noising algorithm was compiled using MATLAB Version 7.11.0.584(R2010b) on a computer system with Intel ® Core™ i7CPU (Q740 @ 1.73 GHz), and 4.00 GB RAM memory.

For mineral density calibration, the grey level values of 10 points on selected images of each HA phantom, were measured, averaged and plotted against mineral density value of the phantoms in order to calculate the calibration equation which was as follows:

Mineral density = (Grey level value × 0.0159) + 0.2967

The calibration equation was used to transform the grey level values of the images into true mineral density values. Each lesion was color coded based on the grey level values and corresponding mineral densities to provide a mineral map of the lesion and render an enhanced visualization. The visualization and color coding were performed using colormapeditor command by choosing Jet color map in MATLAB. Fixed RGB (Red, Green, Blue) index values were set for all the colorized images.

The values of lesion parameters including the mineral density and the thickness of the surface layer were extracted from mineral density profiles plotted in FIJI and the visualized mineral maps in MATLAB.

For calculating the mineral density (MD) of each region, the MD values of all of the pixels in the region were measured and averaged. In the carious lesions, the regions of interest (ROIs) were selected at the surface layer as well as the center of the lesion body. In the sound specimens, the ROIs were selected at the outer layer as well as the inner layer of enamel.

The thickness of the surface layer of the carious lesions was measured using line scans of the mineral content profile across the sidewalls and the floor of the fissures and at the central traverse of the proximal lesions. The distance from the enamel surface up to the first reduction step in the mineral density profile indicating the beginning of the lesion body was measured on the density profiles. In cases where there was not an obvious reduction step in the mineral profile, the value was obtained by measuring the thickness of the color zone corresponding to the surface layer in the colorized images. The measurements were checked for normal distribution. The statistical analysis was performed using statistical software GraphPad Prism (Graphpad Software, San Diego, CA) to test for differences between the means. The results were analyzed by One-way analysis of variance (ANOVA). Multiple comparisons between groups were performed by posthoc Tukey test. P -values less than 0.05 were considered to be statistically significant.

Only gold members can continue reading. Log In or Register to continue

Jun 19, 2018 | Posted by in General Dentistry | Comments Off on Quantitative characterization and micro-CT mineral mapping of natural fissural enamel lesions
Premium Wordpress Themes by UFO Themes